Influence
Part 11  The Defense Codex
Chapter 318 of 360

Fifty-Four Percent: Why You Cannot Spot a Liar and What Works Instead

Bond and DePaulo's meta-analysis is the number this entire guide is organized around: 54 percent accuracy against a 50 percent baseline, across roughly two hundred studies. Professionals score the same as students. Individual differences are negligible. Confidence is uncorrelated with accuracy.

The defensive instruction that follows is not to try harder. It is to stop.

Passive assessment of whether someone is lying is a coin flip with extra steps, and the cues most people use — gaze aversion, fidgeting, nervousness, hesitation — have no validity and correlate instead with anxiety, neurotype, culture and trauma history. Using them makes your judgment worse than not using them, which is chapter 52's conclusion.

What works operates on the account rather than on the person, and there are three methods with real evidence.

Unanticipated questions. Liars prepare, and preparation produces fluency. Ask what nobody rehearses: spatial layout, who was sitting where, what was to your left, what the room smelled like, draw it. Genuine episodic memory contains perceptual and spatial detail as a matter of course. A constructed narrative does not, because nobody rehearses the floor plan.

Reverse order. Vrij's 2008 study, chapter 108, found that asking for an account backwards raised observers' detection accuracy substantially — not by making observers better but by making liars worse. A fabricated account rests on a chronological script, and running it backwards destroys the scaffolding.

Strategic use of evidence. Chapter 100. Withhold what you already know, obtain a full account first, then compare. What you get is not an impression but an artifact: a discrepancy between a statement and a document, which anyone can examine. Hartwig and Granhag's trained interviewers reached accuracy in the region of eighty-five percent, against chance for those who disclosed evidence early.

The common property is the point. None of these assess the person. All of them produce something checkable.

The early signal to watch for in yourself is the feeling of certainty. Confidence in this domain is uncorrelated with accuracy, which means a strong intuition that someone is lying is not evidence about them — it is a report about your own state, and Part 2 documents what happens when institutions act on it.

The counter-response is to replace is he lying with three questions that have answers: what in this account can be independently checked, what would I expect to find if it were true, and what has he told me that conflicts with something I already know and have not mentioned.

This counters Law 27.

The case

Bond and DePaulo’s meta-analysis of 206 deception-judgment studies (Personality and Social Psychology Review, 2006), finding 54% average accuracy, alongside Vrij’s cognitive-load approach and Granhag and Hartwig’s Strategic Use of Evidence technique

The mechanism

Bond and DePaulo established that unaided human lie detection barely beats chance, and that confidence is uncorrelated with accuracy — the intuitive ‘tells’ most people trust are noise. Vrij’s cognitive approach works instead by increasing mental load (unexpected questions, reverse-order recall) so that lying, which is cognitively costlier, degrades measurably. The SUE technique withholds evidence until the suspect commits to a story, converting inconsistency rather than nervousness into the diagnostic signal.

What this chapter covers

  1. The Threat Pattern: False Confidence in Gut Reads
  2. Early Warning Signals & Physiological Tells
  3. Verified Detection Case: Bond & DePaulo’s 206-Study Meta-Analysis
  4. Detection Protocol: Unexpected Questions, Reverse Order, Withheld Evidence
  5. Counter-Response: Test Claims, Not Demeanor

Counters Law 27, Confess the Small Sin to Bury the Large One